Frequency estimation algorithm based on local polynomial approximation

In this paper, we present a novel non-parametric frequency estimation algorithm designed for use in electric power system control and protection applications. This algorithm uses locally a truncated Taylor expansion in order to approximate a time-varying phase angle. This truncated Taylor series containing first derivative of the phase angle (frequency) is fitted locally by a least squares problem. The central point of this algorithm, the way of window size selection for local approximations, is described in detail in this paper. An example is presented at the end of the paper that clearly demonstrates the main features of this method.